A 7 year record of above-ground net primary production in a southeastern Mexican mangrove forest

A 7 year record of above-ground net primary production in a southeastern Mexican mangrove forest

Aquatic botany ELSEVIER Aquatic Botany 55 (1996) 39-60 A 7 year record of above-ground net primary production in a southeastern Mexican mangrove for...

1MB Sizes 0 Downloads 71 Views

Aquatic botany ELSEVIER

Aquatic Botany 55 (1996) 39-60

A 7 year record of above-ground net primary production in a southeastern Mexican mangrove forest J.W. Day Jr. a,b,*, Carlos Coronado-Molina c F.R. Vera-Herrera c R. Twilley o, V.H. Rivera-Monroy ~, H. Alvarez-Guillen c, R. Day e, W. Conner f a Department of Oceanography and Coastal Sciences and Coastal Ecology Institute, Louisiana State University, Baton Rouge, LA 70803, USA b EPOMEX Program, Universidad Autdnoma de Campeche, Campeche, M£xico c Sede 'El Carmen', Instituto de Ciencias del Mar y Limnologfa, Universidad Nacional A ut6tu~ma de Mdxico, Cd. del Carmen, Campeche, M~xico d Department of Biology, P.O. Box 42451, University of Southwestern Louisiana, Lafayette, LA 70504, USA e Southern Science Center, National Biological Service, 700 Cajundome Blvd., Lafayette, LA 70506, USA f Belle Baruch Forest Institute, P.O. Box 596, Clemson University, Georgetown, SC 29442, USA

Accepted 22 April 1996

Abstract Spatial and temporal variations in net above-ground primary production (NPP) and litter turnover rate were studied, from 1987 to 1993, in a mangrove forest bordering Laguna de Terminos, Mexico. NPP, the sum of total litter fall and wood production, was measured over the entire study period in three zones in a basin forest: zone I, where Rhizophora mangle (red mangrove) occurs but Avicennia germinans (black mangrove) is the dominant species; zone |I, a scrub forest of A. germinans; zone III, where larger A. germinans trees occur. In 1991, a fringe zone dominated by A. germinans and R. mangle was added to the study. Three distinctive climatic seasons occur in the region: rainy, dry, and cold front (locally named 'nortes'). Average total litter fall in the fringe zone (793 g m - 2 year- i) was significantly higher than in the basin forest (496, 307, and 410 g m -~ year-1 for basin zones I, II, and III, respectively). All zones showed significant differences among seasons with the norte season having significantly lower litter fall. Litter turnover rates were about 7 months in zones I and II and 10 months in zone Ill, reflecting the low tidal range that occurs in the basin forest. Low litter turnover rates in the basin

* Corresponding author. 0304-3770/96/$15.00 Copyright © 1996 Elsevier Science B.V. All rights reserved. PII S0304- 3770(96)01063-7

40

J.W. Day Jr. et al./ Aquatic Botany 55 (1996) 39-60

forest were reflected in a high organic matter standing crop. Annual average stem growth was significantly higher in zones I and III (1.27 and 1.36 kg per tree year-1, respectively) than in zone II (0.62 kg per tree year-l). Above-ground NPP rates in the basin forest (399-695 g m -2 year - l ) were lower than in fringe and riverine forests, reflecting patterns of litter fall and wood production. There was no seasonal variation in soil salinity but the basin forest had significantly higher soil salinity than the fringe forest. Spatially, mean soil salinity was inversely related to litter fall. Long-term patterns in soil salinity, precipitation and air temperature explained 74% of the inter-annual litter fall variability. Over the 7 year study, productivity in zone II was more variable than in zones I and III, and productivity (litter fall and wood growth) were less variable than litter standing crop and turnover. Keywords: Terminos Lagoon; Litter fall; Long-term productivity

1. Introduction A number of factors affect the structure and productivity of mangrove forests including hydrology, nutrient inputs, soil salinity, and soil type (Lugo et al., 1988). Lugo and Snedaker (1974) developed a mangrove classification based on differences among these factors which includes five basic types of mangrove forests: riverine, fringe, overwash, basin, and dwarf. More recently, the five types have been reduced to three basic forest types: basin, fringe, and riverine (Twilley et al., 1986; Lugo et al., 1988). Characteristics such as structure, range of productivity, litter decomposition, and detritus export, along with hydrologic and nutrient factors, define each type of forest in such a way that total biomass, tree height, litter fall and decomposition, and freshwater turnover increase from basin to riverine forests (Pool et al., 1977; Lugo et al., 1988). For the majority of mangrove productivity studies, primary productivity (mainly litter fall) has been measured over a short time period, usually 1 year. Even though mean annual litter fall increases from basin to fringe to riverine forests (Lugo et al., 1988), there is considerable variability within each class. For example, reported total annual litter fall ranges from 320 to 1700 g m -2 year-1 for riverine forests, 430-1082 g m -2 year-1 for fringe forests, and 250-970 g m - 2 y e a r - l for basin forests (Lugo et al., 1988). Within basin forests, monospecific Avicennia sp. forests have lower productivity than mixed species forests. However, it is not always clear if site to site variability is due to site specific characteristics or to inter-annual changes in such factors as rainfall, river discharge, temperature, or salinity. To resolve this question, long-term studies will help determine the degree to which ranges in reported mangrove productivity are due to inter-annual variability. In this study, we examined 7 years of data to help clarify factors affecting long-term patterns of litter production. In this study, we analyzed net above-ground primary production of fringe and basin mangrove forests bordering Laguna de Terminos. The study had two main objectives: (1) to describe the spatial and seasonal patterns of litter fall production in both monospecific basin and mixed fringe mangrove forests, and (2) to assess the influence of several environmental factors, such as interstitial water salinity, rainfall and air temperature, affecting the inter-annual variability of mangrove productivity.

J.W. Day Jr. et al./Aquatic Botany 55 (1996) 39-60

41

2. Materials and methods 2.1. Description of area Laguna de Terminos is a large (approximately 1800 km2), shallow (mean depth 3.5 m) coastal lagoon bordering the southern Gulf of Mexico in the state of Campeche, Mexico (18040 ' N, 91 °80' W). It is connected to the Gulf of Mexico via two inlets located at each end of Isla del Carmen, a barrier island that separates the lagoon from the open gulf. The lagoon has a net east to west flow caused by prevailing trade winds. The mean tidal range is about 0.5 m. The highest riverine input, which has a marked effect on the salinity and hydrology of the lagoon water, occurs during October and November (Y(uSez-Arancibia and Day, 1982). The area has a humid, tropical climate. Average annual precipitation (1680 mm) is seasonal with low rainfall occurring from February to May and high rainfall from July to November. The same pattern exists over the entire drainage basin of the lagoon, producing a similar pattern river flow into the lagoon. Based on patterns of rainfall, air temperature and wind, three climatic seasons have been described for the area: the dry season from mid February to early June, the rainy season from June through October, and the 'norte' or cold front season occurring from November to February. Mean monthly air temperature ranges from 18 to 36°C with the norte season having the lowest, and the rainy season the highest, temperatures. Maximum wind speeds (50-80 km h- t) occur during the norte season. The lagoon is bordered almost completely by extensive mangrove swamps which extend up the rivers and associated small lagoons to the limit of tidal influence. Three tree species dominate: Rhizophora mangle L. (red mangrove), Avicennia germinans L. (black mangrove), and Laguncularia racemosa Gaertn. f. (white mangrove). A fourth species, Conocarpus erecta L. (buttonwood), grows only along the landward fringes of the swamp. Beds of tropical sea grasses occur along the lagoon side of Carmen Island and along the eastem and southeastern shores of the lagoon. This distribution clearly reflects patterns of circulation, water transparency, and salinity. A more detailed description of the area is given by Yfifiez-Arancibia and Day (1982). The study was conducted in a mangrove forest bordering Estero Pargo, a tidal channel 5.9 km long and averaging 80 m in width, located on the lagoon side of Carmen Island. There is a regularly inundated fringe forest zone, about 20 m wide, adjacent to the channel. This zone is dominated by R. mangle although A. germinans and L. racemosa also occur. Behind the fringe forests there are infrequently flooded basin mangroves dominated mainly by A. germinans. The fringe forest drains completely on each tide. The basin forest, however, is continuously flooded during most of the wet season and. is often dry during the dry season, indicating that there is little groundwater flow (Rivera-Monroy et al., 1995). Preliminary observations of the basin forest indicated that there were three distinct zones. The first zone (Zone I) is located 25-60 m inland from the channel and is characterized by a mixed forest dominated by A. germinans. In this zone, average tree height is 6 - 7 m and mean dbh is 7.3 cm. The second zone (Zone II) is located 60-120 m landward from the creek and is composed of small A. germinans trees of 3 - 4 m

42

J.W. Day Jr. et al. /Aquatic Botany55 (1996)39-60

average height and 4.8 cm mean diameter at breast height (dbh). The third zone (Zone III) is located 120-180 m inland, and is composed of A. germinans with an average height of 6 - 7 m and mean dbh of 7.8 cm. Owing to economic and logistical limitations, it was impossible to randomly sample the entire statistical population of interest, i.e. the entire extent of the several zones. Based on visual inspection of the area and the opinion of the investigators, a transect was established in a representative area of the mangrove forest. We assume that the statistical distribution of the responses measured in and around the transect is the same as in the statistical population of interest. In 1987, two 50 m × 20 m plots were randomly established in each zone where each plot was considered to be a replicate within a zone. Tree density, height, and canopy cover are well developed in zones I and III and understory vegetation is poorly developed. In zone II the canopy cover is low and Salicornia spp. occurs. In 1991, two additional plots, for litter fall and structural analysis, were added in the fringe forest zone adjacent to Estero Pargo, where each plot was considered to be a replicate within that zone. Avicennia germinans dominates in this zone, but L. racemosa and R. mangle also occur. Trees in this zone have an average height of 8 - 9 m and a mean dbh of 8.1 cm. The structure of the forest has not been significantly affected by logging or hurricanes. A detailed description of the study area at Estero Pargo is given by Rivera-Monroy et al. (1995). 2.2. Litter fall and standing crop measurements

Five 0.25 m 2 litter collectors, made of wood frames lined with plastic screening (1 mm mesh), were randomly set up in each plot located within the basin and fringe forests above the level of the highest tide. From January 1987 until December 1993 litter fall was collected approximately monthly. Litter from each collector was oven-dried to constant weight at 70°C, sorted by species into leaves and non-leaf material (fruits, flowers, twigs, and bark), and weighed. Litter standing stock was sampled in the basin forest plots four times during the first 2 years and then once every climatic season during the rest of the study period. Five O. 1 m 2 sites were randomly chosen in each of the six basin forest plots for collecting litter from the surface of the forest floor (L horizon) in the vicinity of the litter fall collectors. Standing crop samples were processed in the same manner as litter fall. The residence time of litter on the floor of the basin forest was estimated using the equation 1 / k = L / X s s developed by Olson (1963), where L is the input (litter fall production), Xss is the steady-state content of litter accumulated on the floor forest (standing crop), and 1 / k is the turnover rate or residence time in years. When litter accumulation (Xss) reaches a steady-state, the rate of change ( L / X s s ) is zero, indicating that input is equal to loss. 2.3. Forest structure, annual stem production and total above-ground primary production

In 1987, dbh and height of all trees over 2.5 cm in dbh were measured in each of the six basin forest plots in 1987 and in the fringe plots in 1991. From these data, basal area, average diameter, stem density, mean canopy height, and complexity index were

J.W. Day Jr. et al./Aquatic Botany 55 (1996) 39-60

43

calculated (Pool et al., 1977; Day et al., 1987; Lugo et al., 1988). Then 25 A. germinans trees per basin plot were randomly chosen, tagged, and their dbh measured. During the study period, dbh of the tagged trees was measured once a year at the beginning of the dry season (usually in February). The biomass of each tagged tree was estimated using the regression equation logeY= - 1 . 5 8 5 2 + 2.302 loge (x) formulated by Day et al. (1987) for A. germinans, where Y is dry weight (kg) and x is dbh (cm). The increase in woody biomass per tree for each year was computed by subtracting the initial biomass from the final biomass. Initial diameter was then plotted against the increase in biomass per tree and regression equations obtained so that for any diameter it was possible to predict the biomass increase. The equations used for predicting biomass increase during the study period were as follows 1987-1988 1988-1989 1989-199(I 1990-1991 1991-1992 1992-1993 1993-1994

Y= Y-= Y= Y= Y= Y= Y=

- 1.049 + -0.985 + -0.408 + -0.436 + -0.419 + - 1.241 + -0.908 +

0.347 0.316 0.246 0.274 0.353 0.365 0.244

(x) (x) (x) (x) (x) (x) (x)

R2= R2= R2= R2= R2= R2 = R2=

0.72 0.59 0.48 0.46 0.40 0.50 0.40

(P (P (P (P (P (P (P

< < < < < < <

0.05) 0.05) 0.05) 0.05) 0.05) 0.05) 0.05)

where Y is the predicted biomass increase and (x) is dbh. Diameter at breast height data were grouped by 2.5-cm-diameter classes and the number of trees per hectare in a given diameter class was then multiplied by the predicted biomass increase for that diameter class. The sum of the biomass increases for all diameter classes gave the total net wood production, in g m -2 year- ~, of the mangroves in each zone. Since A. germinans trees are dominant at the plots located in the basin forest, biomass increase was estimated only for this species. Litter fall and stem production for each year were summed to obtain estimates of annual net above-ground primary production. Ratios of stem production to total litter fall were calculated for each zone for each year.

2.4. Interstitial water salinity and climate data Three transects, with ten groundwater wells each, were set up perpendicular to the tidal channel. The transect extended 210 m through all three zones. The wells, spaced at 20 m intervals, were placed into the soil to a depth of 0.5 m. The bottom 30 cm of the wells were perforated and covered with fine screening to allow water exchange but prevent entry of soil material. The wells extended above the soil surface to a level higher than the highest tide and were covered to prevent entry of rainwater and evaporation between samplings. Interstitial water salinity samples were pumped from each well monthly during the study period and measured using an optical refractometer. Daily readings of temperature and rainfall were taken at the marine laboratory of the National University of Mexico which is located adjacent to the study site. Monthly and seasonal averages of these data were calculated from the daily readings. Average soil salinity for each zone during each sampling period was calculated using all wells located in that zone.

44

J.W. Day Jr. et al./Aquatic Botany 55 (1996) 39-60

2.5. Statistical analyses Litter fall, stem production, above-ground net production and interstitial water salinity were used for testing for significant differences ( P < 0.05) among zones and climatic seasons with an ANOVA split-plot analysis. The main plot tested the factor zone (I, II, and III); the subplot tested climatic seasons (dry, rainy, and norte) and the interaction Zone × Season. After performing the analyses, linear contrasts were used to assess differences between treatments. Backward and stepwise multiple regression analyses were used to relate litter fall production values to average, maximum and minimum records of precipitation, air temperature and interstitial water salinity. Violations of the assumptions for the method were checked by residual analysis. An ANOVA was performed to test if there were differences between years for the stem production to litter fall ratio. A Shapiro-Wilk test was performed to determine whether the residuals of the ratios were normally distributed. The yearly wood-litter ratio for each zone was regressed separately against mean annual rainfall, temperature, and soil salinity to determine if there were relationships among these variables. Statistical analyses were run with JMP ® (Statistical Analysis Systems Institute Inc., 1991). 3. Results

3.1. Temporal and spatial patterns of litter fall Annual mean leaf fall rates for individual years for the three basin forest zones ranged from 0.8 to 1.25 g m -2 day -I. The mean annual leaf fall pattern for the entire study period in the basin forest was bimodal, generally with a higher peak at the end of the dry season (May) and a slightly lower peak during the rainy season (October) (Fig. 1). Average leaf fall for the entire study was lowest during the norte season. Mean monthly rates of the fall of non-leaf material for the overall study ranged from 0.06 to •

BASIN NGE

"o

E .A ii tll

J

F

M

A

M

J

J

A

S

0

N

D

TIME (MONTH) Fig. I. The annual pattern of leaf litter fall in basin and fringe mangrove forests. Each bar is the average for that month over the entire study period. Lines show one standard error.

J.W. Day Jr. et al./Aquatic Botany 55 (1996) 39-60

45

0.2 g m -2 day-~ and showed no clear seasonality, although higher rates were generally observed during the rainy and dry seasons. Annual mean leaf fall for individual years accounted for 86-95% of the total litter fall. Also, during the whole study period, A. germinans contributed about 98% of the total litter fall while L. racemosa and R. mangle contributed the remaining 2%. For this reason, the analyses reflect primarily A. germinans data (although all litter fall data were included). Average leaf fall rates over the entire study in the mixed fringe forest (Fig. 1) ranged from 1.09 (norte season) to 2.91 (dry season) g m -2 day-~. The long-term average pattern was similar to that of the basin forest, with the driest months having the highest leaf fall production and the lowest values occurring during the norte season. Fall of non-leaf material seasonal pattern during individual years was highly variable, ranging from 0 during the norte season to a high of 0.47 g m - 2 d a y - ~ during the rainy season. During the whole study period, leaf fall contributed 88-95% of the total annual litter fall in the fringe forest. These proportions are similar to that observed in the basin forest, and since leaf fall contribution to litter fall was high, total average litter fall seasonal pattern over the entire study period was similar to that of the leaf fall. In the mixed fringe forest, leaf fall contribution for each species was 53%, 42%, and 5% for R. mangle, A. germinans, and L. racemosa, respectively. Spatially, annual average total litter fall ranged from 0.59 g m -2 day-~ for zone II in 1989 to 1.51 g m -2 day -~ for zone III in 1993. The annual average litter fall for the 7 year period was 1.34, 0.83, and 1.13 g m -2 day-~ for zones I, II, and III, respectively, with zone II having the lowest litter fall rates during the entire study period (Fig. 2a). Over the 7 year period, total litter fall varied by a factor of 1.5-1.8 within each zone. The coefficients of variation of annual total litter fall over the study period were 13.6%, 23.3%, and 20.0% for zones I, II, and III, respectively. Seasonally, overall mean total litter fall in the basin forest ranged from 0.94 (norte season) to 1.37 (dry season) g m 2 day-], The seasonal pattern was similar to that of the leaf fall with dry and rainy seasons having higher litter fall rates than the norte season (Fig. 2b). The average overall total litter fall was 1.28, 1.23, and 0.85 g m -z day-J for dry, rainy, and norte seasons, respectively. Statistical analyses showed no spatial differences. In contrast, the analysis showed highly significant differences by season. Linear contrast analysis within each year showed that litter fall production during the norte season was significantly lower than that during the dry and rainy seasons ( P < 0.05). Total lilter fall rates in the fringe forest were, spatially and seasonally, significantly higher than that of the three basin forest zones during the 3 years of comparable study (Fig. 2a). Linear contrast analysis within each year confirmed that litter fall rates in the basin zones were similar among each other and that they were significantly different from the fringe area ( P < 0.05). Moreover, the contrast showed that litter fall production during the dry and rainy seasons was not significantly different and that litter fall was significantly lower during the norte season. 3.2. Litter standing crop and turnover rates

Annual litter standing crop ranged from 60 to 477 g m - 2 , with a mean of 156 g m -~' During the whole study period, annual average standing crop was highly variable,

46

J.W. Day Jr. et al./Aquatic Botany 55 (1996) 39-60 (a)

3

m

.....

"ID

~,

E

2 ¸

ID~

v -J --I Q:

UJ

1

..J

87

88

89

90

9~

92

93

91

92

93

TIME (YEAR) (b)

2-



DRY SEASON

[]

RAINY SEASON

[] NORTE SEASON

E -1

1

U.I

.J

o

87

88

89

90

TIME (YEAR) Fig. 2. Inter-annual variation of total litter fall: (a) by zone and (b) the mean by season of both basin and fringe mangrove forests. Each bar is the mean for each year for the respective zone or season. Lines show one standard error.

ranging from a low of 60 g m -2 in zone II in 1989 to 425 g m -2 in zone I in 1987 and 439 g m -2 in zone III in 1992. The coefficients of variation of annual average litter standing crop over the study period were 51.5%, 43.6%, and 44.4% for zones I, II, and III, respectively. Mean annual values in the basin forest plots were 261, 151, and 303 g m -2 for zones I, II, and III, respectively (Table 1). Statistical analysis testing spatial and temporal differences showed significant differences among zones and years ( P < 0.05). Contrast analysis within each year showed that litter in zone II was significantly lower ( P < 0.05) than the other two zones. The annual average residence time of the litter standing crop over the 7 year study was 203 days (6.7 months) in zone I, 227 days (7.6 months) in zone II, and 302 days

47

J.W. Day Jr. et al./Aquatic Botany 55 (1996) 39-60

Table 1 Annual average of total litter standing crop over the entire study period Zone 1 I1 III

Totallitterstanding crop(g m -2 ) 1987

1988

1989

1990

1991

1992

Mean

425 t86 420

316 181 393

119 60 128

160 127 167

151 109 271

396 246 439

261 151 303

(10.0 months) in zone III (Table 2) but these differences were not significant at P < 0.05. By contrast, there were significant differences in inter-annual litter turnover with 1989, 1990, and 1991 having the lowest rates. The coefficients of variation of annual average litter turnover over the study period were 56.1%, 40.1%, and 45.8% for zones I, II, and III, respectively. 3.3. Forest structure, biomass and annual stem production

Zone II had lower mean dbh (4.9 cm), basal area (3.8 m 2 h a - J ) , and mean canopy height (4 m), resulting in a lower complexity index (2.8), while zones I and III had higher mean dbh (7.3 and 7.7 cm, respectively), basal area (10.4 and 11.9 m e h a - ~), and mean canopy height (6 m) (Table 3). As a result, zones I and III had higher complexity indices (21.2 and 11.4, respectively). The fringe forest had a mean dbh of 8.8 cm, a basal area of 26.3 m 2 ha - l , a mean canopy height o f 8 m and a complexity index of 22.9. Average biomass of A. germinans was 5.78, 1.74, and 7.30 kg m -~ for zones I, II, and III, respectively, compared with 16.4 kg m -2 for the fringe zone. The higher biomass estimated in the fringe zone is related to the high dbh (mean 8.8 cm) and canopy height (mean 8 m) observed in this zone. During the entire study period the average wood production in the basin forest (Table 4) ranged from 0.28 to 1.92 kg per tree y e a r - ~. On a temporal and a spatial basis, zone II showed the lowest growth rates, ranging from 0.28 kg per tree y e a r - i in 1993 to 0.81 kg per tree y e a r - ~ in 1989; zone I had the highest rates, varying from 1.92 to 1.02 kg per tree year -~, yet temporal variability was not significant. Spatial variability during the whole study period was statistically significant where average growth rates were lower in zone II (mean 0.62 kg per tree y e a r - ~) and higher in zones I and III (mean 1.27 and 1.36 kg per tree y e a r - J,

Table 2 Annual average of litter turnover rates in the basin forest Zone

Littertumover(days) 1987

1988

1989

1990

1991

1992

Mean

I 11 11I

367 358 420

254 272 410

96 128 154

123 212 167

96 124 212

280 268 450

203 227 302

48

J.W. Day Jr. et aL //Aquatic Botany 55 (1996) 39-60

80

~ ' b-

•=


~

,~

o

m

m

¢-

•~ ~ ° ~

A

C~

•-~

~

~

i

,n

b o

J.W. Day Jr. et al./Aquatic Botany 55 (1996) 39-60

49

Table 4 Wood production annual average of Avicennia germinans in the basin mangrove forest Zone

I 11 II1

Wood production (kg per tree year i ) 1987

1988

1989

1990

1991

1992

1993

Mean

1.61 0.68 1.64

1.33 0.58 1.46

1.42 0.84 1.52

1.22 0.71 1.22

1.03 0.63 1.11

1.41 0.57 1.58

0.88 0.31 0.99

1.27 0.62 1.36

respectively). The coefficients of variation of annual wood production per tree over the study period were 19.6%, 26.5%, and 18.5% for zones I, II, and III, respectively. Annual average total wood production had the same temporal and spatial pattern as the individual tree growth rates, with zone II having the lowest wood production, ranging from 49.8 to 147.2 g m -2 year -1, and zones I and III ranging from 130.9 to 239.1 g m -2 year -1 and 144.1-261.2 g m -2 year -~, respectively (Table 5). The

Table 5 Estimations of above-ground net primary production in the basin mangrove forest Component

Zone 1 Annual average wood production Avicennia germinans Litter fall Leaves Wood Total Annual above-ground net primary production Zone H Annual average wood production Avicennia germinans Litter fall Leaves Wood Total Annual above-ground net primary production Zone 111 Annual average wood production Avicenniagerminans Litter fall Leaves Wood Total Annual above-ground net primary production

Above-ground net primary production (g m - z year- ~) 1987

1988

1989

1990

1991

1992

1993

239.1

211.1

224.9

206.1

1 3 8 . 8 224.5

130.9

361.1 60.2 421.6 660.7

423.4 31.0 454.4 665.5

408.8 60.2 469.0 693.9

439.8 21.9 461.7 667.8

500.1 120.4 620.5 759.3

458.1 54.8 5 l 2.9 737.4

489.1 58.4 547.5 678.4

121.4

t 04.0

147.2

104.7

130.6

I 01.1

49.8

222.6 18.2 240.8 362.2

266.4 21.0 287.4 319.4

204.4 11.5 215.9 363. I

240.9 6.0 246.9 351.6

333.9 20.1 354.0 484.6

361.4 5.5 366.9 470.1

381.4 10.6 392.0 441.8

261.2

222.3

245.6

223.4

144.1

147.3

145.6

317.6 49.3 366.9 628.1

350.4 29.2 379.6 601.9

266.4 67.5 339.9 585.5

354.1 29.2 383.3 606.7

443.5 78.5 522.0 666. I

350.4 12.4 362.8 510.1

485.4 58.4 543.8 689.4

50

J.W. Day Jr. et al./ Aquatic Botany 55 (1996) 39-60

coefficients of variation of annual average litter wood production over the study period were 22.1%, 28.3%, and 25.8% for zones I, II, and III, respectively. 3.4. Total above-ground net primary production

Annual above-ground NPP ranged from 319.4 to 759.3 g m -2 year-1 with zone II having the lowest values, ranging from 319.4 to 484.6 g m -2 year-1 and zones I and III having the highest values varying from 660.7 to 759.3 g m -2 year- l (6.61-7.59 t hayear - i ) and 510.1-689.4 g m -2 year - l , respectively (Table 5). Even though total above-ground production in zone II was lower than in zones I and III, there were no significant differences ( P < 0.05) among areas. By contrast, inter-annual variability was significantly different with 1991, 1992, and 1993 having higher total above-ground production than the four previous years. The coefficients of variation of annual N-PP over the study period were 5.5%, 16.4%, and 9.5% for zones I, II, and III, respectively. The stem production to litter fall ratio for the basin forest ranged from a low of 0.13 for zone II in 1993 to a high of 0.72 for zone III in 1989 and the overall mean was 0.41. There were significant differences between years ( P > F = 0.0009). The regression of salinity vs. the ratio explained 54% of the variation and was the only significant predictor ( P < 0.05). 3.5. Interstitial water salinity

Spatially, interstitial salinity increased from the fringe forest to zone III of the basin forest (Fig. 3). The annual mean soil salinity in the basin forest was 57.3, 65.5, and 70.2 ppt for zones I, II, and III, respectively, while in the fringe forest was 46.4 ppt. The spatial pattern observed during the whole study period, low in the fringe forest and high in the basin forest, was statistically significant ( P < 0.05). Linear contrast analyses confirmed that there were spatial differences and showed that the fringe forest was

80

~

m

6O

50

40

Fringe Zone

Zone III

/ ,

0

.

Pond

,

20

.

,

40



,

60

|

80

i00

-

i

120



,

140



,

160



,

180

DISTANCE FROM TIDAL CHANNEL (m)

Fig. 3. The interstitial salinity gradient through both fringe and basin forests. Each point is the averagefor that location over the entire study period. Vertical lines show + one standard error.

J.W. Day Jr. et al./Aquatic Botany 55 (1996) 39-60 2.5

q

E

y

=

3.9176

-

0.04

x

R^2

:

51

0.77

2.0

,-1 1.5 ILl .-1 .-1 ~

o

1.0

0.5 40

J

i

i

i

50

60

70

80

SOIL SALINITY (ppt)

Fig. 4. Regression between total litter fall and soil salinity for fringe and basin zones for 1991, 1992, and 1993. significantly lower ( P < 0.05) than the inland basin zones and that the three basin areas were not significantly different. Temporally, mean interstitial salinity did not vary significantly and ranged from 59.8 to 72.4 ppt. Mean salinity values for each climatic season were 64.3 ppt (Dry), 65.8 ppt (Rainy), and 65.9 ppt (Norte). Linear contrast analysis confirmed, also, that different climatic seasons were not significantly different, indicating that seasonal variation in rainfall was not reflected in soil salinity changes. Tidal activity has been shown to be an important factor influencing soil salinity variability, as has been reported for Rookery Bay (Twilley et al., 1986). Spatially, annual average litter fall in the basin and fringe zones for 1991, 1992, and 1993 was inversely related to annual average soil salinity (Fig. 4) according to the following equation Y --- 3.915 - 0.039( x)

R 2=0.77(P<0.001)

where Y is annual average litter fall (g m -2 year-~) and x is annual average soil salinity (ppt).

3.6. Inter-annual productivity patterns There were clear seasonal variations in both rainfall and air temperature over the 7 years of the study. Seasonally, precipitation during the rainy season averaged 220.9 _+ 31.1 mm (for the five rainy months) and during the dry season it averaged 35.6 +_ 29.8 mm during March, April and May. Precipitation during the norte season averaged 106.6 _+ 42.1 mm per 4 months. Seasonal air temperature pattern was regular and ranged from 24.1 __%1.1°C in the norte season to 27.9 _+ 0.7°C in the dry season and 29.3 _+ 2.1°C in the rainy season. Both backward and stepwise multiple regression analysis selected

J. W. Day Jr. et al. /Aquatic Botany 55 (1996) 39-60

52

1.8!

o

Litterfail I

(a)

[ - * - - Soil Salinity

1.6~ ~

75 -70 ~

1.4

f

-6s ~.~ r~

-6o N

0.8: 0.6"

55

DRN

ta

~

DRN

DRN

Litterfall ]

DRN

DRN

(b)

DRN

DRN

[ -*-- AirTemp

1.8

30

1.6-

-28 ~--

1.41.2-

~

1-

~-

0.80.6

,", .... DRN

0.6

, .... DRN DRN

Litterfail [

1.8

1.4-

/,

~

• DRN 87

,

DRN 88

',I\

-24

i .... i .... DRN DRN DRN

-22 [~ .~. ,< 2O

i. DRN

(¢)

[ - * . - - Rainfall

~}'

~

.

DRN 89

-26

.

DRN 90

.

.

DRN 91

.

10o

0

.

DRN 92

200

DRN 93

Fig. 5' T h e r e l a t i o n s h i p b e t w e e n l o n g - t e r m s e a s o n a l p a t t e r n s o f total litter fall a n d (a) soil salinity, (b) t e m p e r a t u r e , a n d (c) rainfall. D is d r y s e a s o n , N is n o r t e s e a s o n , a n d R is r a i n y s e a s o n .

average soil salinity (Fig. 5a), minimum air temperature (Fig. 5b), and minimum precipitation (Fig. 5c) as independent variables which were highly significant in explaining temporal litter fall variability over the 7 year study. The equation that best describes litter fall variability is Y=-3.517+0.024(Xt)+0.130(X2)-0.004(X3)

R2 = 0 . 7 4

where X l is average soil salinity (P < 0.05); X2 is minimum temperature (P < 0.01) and X 3 is minimum rainfall (P < 0.01).

4. Discussion

The factors which control the structure and productivity of mangrove forests vary in time and space. Inter-annual variability in mangrove productivity is affected by long-term

J.W. Day Jr. et al. /Aquatic Botany 55 (1996) 3 9 - 6 0

53

variation in such factors as mean sea level, air temperature, river flow, total precipitation and soil salinity, which is affected by the first four factors (Williams et al., 1981; Thorn, 1982). These factors are the result of forces which act over broad areas and should produce similar inter-annual patterns of productivity in different mangrove forests in the same geographic region (such as the forests in the Laguna de Terminos system). Seasonal patterns of production respond to seasonal changes in rainfall, river flow, soil salinity, air temperature, and irradiation (Clough and Attiwill, 1982; Day et al., 1987; Flores-Verdugo et al., 1987; Clough, 1993). Spatially, mangroves respond to differences in freshwater inflow, nutrient inputs, soil salinity, and water turnover rate (Twilley et al., 1986; Robertson and Alongi, 1993). For all of these scales, extremes of the various factors may be more important than means. Here we analyze the role of these factors operating on different time and spatial scales in controlling structure and productivity patterns in the mangrove forests of this system in the neotropics, and compare our results with other long-term studies of mangroves in the old world tropics.

4.1. Seasonal and spatial patterns of litter fall Over the 7 year study, total litter fall was higher in the fringe forest and basin zone II was generally lower than zones I and III. Litter fall rates in the basin forest fall within the lower (351 g m -2 year - t ) to mid range (670 g m -2 year - l ) reported for monospecific basin forests (Christensen, 1978; Goulter and Allaway, 1979; Twilley et al., 1986). Litter fall in the fringe forest was similar to that of another fringe forest in Laguna de T6rminos and other areas around the world (Twilley et al., 1986; Day et al., 1987; Lugo et al., 1988; Robertson and Alongi, 1993). Litter fall was inversely related to soil salinity. High soil salinity has been shown to stress mangroves and cause low litter fall in mangrove forests (Lugo et al., 1988; Twilley et al., 1986). The salinity gradient at Estero Pargo is at least partially due to differential flooding regimes between the fringe forest, which drains completely on each tide, and basin forest, which drains very. slowly. Zones II and III often have standing water for long periods during the rainy seasons (Rivera-Monroy et al., 1995). Studies carried out in Estero Pargo during the same period of this study indicate that nutrient availability is also a likely factor affecting spatial productivity patterns in the mangrove forest at Estero Pargo. Rivera-Monroy et al. (1995) demonstrated that the fringe forest at Estero Pargo is a sink for inorganic nitrogen imported from the tidal creek. In contrast, the basin forest, which is less frequently flooded, receives much lower nutrient inputs and exports nitrogen-poor organic matter to the fringe forest. A number of studies have shown that increased nutrient input is related to higher productivity in mangroves (Onuf et al., 1977; Brown and Lugo, 1982; Boto et al., 1984; Day et al., 1987; Lugo et al., 1988). Thus, better drained soils, lower salinity and regular nutrient input ~re the factors related to higher mean productivity in the fringe forest. The long-term consistency of spatial patterns of litter fall observed at Estero Pargo shows that the hypothesis that mangrove forests with higher water turnover (riverine > fringe > basin) also have higher litter fall rates is a robust one that holds on an inter-annual basis.

J.W. Day Jr. et a l . / Aquatic Botany 55 (1996) 39-60

54

TOTAL LffTERFALL 2-

"o

'7, E I=

1-

Q

J

F

M

A

M

J

J

A

S

O

N

D

TIME

.-B.- MINIMUM RAINFALL -4,-- MEAN SALINITY --

AIR TEMPERATURE

200

30

I 65 -64 -63

.6"

-62

-60 -61 -59

|

i

-58 -57 -56 0

55 J

F

M

A

M

J

J

A

S

O

N

D

Fig. 6, (a) Total litter fall over the 7 year study period. Each bar is the average for that month over the entire study period. Lines show one standard error. (b) Monthly averages of rainfall, temperature and soil salinity. All values are monthly averages over the 7 year study period.

In many mangrove areas, peak litter fall occurs during the rainy season (Leach and Burgin, 1985; Twilley et al., 1986; Day et al., 1987), but Flores-Verdugo et al. (1987) reported high litter fall at the end of the dry season for a small, mangrove system on the Pacific coast of Mexico. From year to year, litter fall at Estero Pargo was highest either in the dry season or during the rainy season and always lowest during the norte season (Fig. 6). This seasonal pattern is similar to that observed in mangrove forests located in the southern Gulf of Mexico in Tabasco (Lopez-Portillo and Ezcurra, 1985), Veracruz (Rico-Gray and Lot, 1983), and in mangrove forests in Papua New Guinea (Leach and Burgin, 1985) where temperature and rainfall were found to be important forcing functions affecting litter fall production. The bimodal litter fall pattern observed at Estero Pargo was similar for both fringe and basin forests, indicating that the entire forest is affected by the same seasonal forcing functions. Seventy-four percent of

J.W. Day Jr. et al. /Aquatic Botany 55 (I996) 39-60

55

seasonal litter fall variability was explained by minimum dry season rainfall, minimum norte season temperature, and mean soil salinity. Other factors such as winds, irradiation, and water level have also been demonstrated to be important factors affecting seasonal litter fall patterns (Lopez-Portillo and Ezcurra, 1985). 4.2. Standing crop and turnover rates The turnover rate of litter on the forest floor in the basin forest was low, in contrast to fringe and riverine mangrove forests where hydrologic energy is high. Litter standing crop in zone II was significantly lower than zones I and lII; thus, the spatial pattern of standing crop reflects the litter fall rates within each zone. Litter fall rates explained 62% of the variation in standing crop ( P < 0.05). Litter standing crop reported in other areas range from 9.9 and 38 g m -2 at Three Fathoms Cove, Hong Kong and Rookery Bay, respectively, to a high of 959 g m -2 at Mai Po, Hong Kong (Twilley et al., 1986; Lee, 1989). There is no tidal inundation at Mai Po while the forest at Three Fathom Cove is strongly flushed (Lee, 1989). High standing litter levels reflect long residence times of litter in areas of minimal tidal activity. The long litter residence time at Estero Pargo reflects decreasing tidal influence and, thus, flushing efficiency across the basin forest. Litter residence times for other mangrove forests show similar patterns which reflect both different flushing rates as well as decomposition differences among mangrove species. For instance, the litter residence time was 6 months at Fort Myers, Florida in a site 250 m from the bay but only 2 months in a site 65 m from the bay (Twilley et al., 1986). The site at Mai Po had litter residence times from 8 to 32 months, reflecting the infrequent tidal inundation at this site (Lee, 1989). Sediments in the basin forest at Estero Pargo are higher in organic matter (42.1%) than in the fringe forest (32.3%) (Lynch et al., 1989). The long residence times in the basin forest cause more of the standing litter to be decomposed and recycled in situ, resulting in the higher organic matter levels. Rivera-Monroy et al. (1995) showed that significant quantities of organic total suspended sediments with a high C / N ratio were exported from the basin forest to the fringe forest at Estero Pargo. This material was then enriched in the fringe forest with inorganic nitrogen imported from the tidal channel to the fringe forest and then exported to the channel as particulate material with a lower C / N ratio. It seems, therefore, that an important ecological role of the basin forest is to provide bulk organic matter to the fringe forest where the nutritional quality is upgraded. 4.3. Patterns of stem production The average growth rates of A. germinans fall within the range of stem growth reported for this genus (Lugo et al., 1988). Avicennia germinans woody growth rate in a riverine forest bordering Terminos lagoon was 2.21 kg per tree year-~ and 1.56 kg per tree year- J in a fringe forest located across the tidal channel from our study site (Day et al., 1987). Total stem production, which was related to average wood production per tree, was similar to values reported for basin forests elsewhere (Lugo et al., 1988). By

56

J.W. Day Jr. et aL /Aquatic Botany 55 (1996) 39-60

comparison, wood production at fringe and riverine sites in Terminos lagoon was 771 and 1205 g m-2 year- 1, respectively (Day et al., 1987). High nutrient inputs and lower soil salinities lead to higher woody growth in riverine forests, compared with fringe and basin forests. The coefficient of variation of growth rates per tree varied by 18-26% and that of annual wood production varied by 22-28% over the 7 year study. Putz and Chan (1986) reported that the average stem production over 31 years for a Rhizophora-Bruguiera forest in Malaysia was 670 g m -2 year-1. The year to year variability was similar to that at Estero Pargo. Average woody biomass increase for A. germinans varies depending on the forest type where this species occurs. For instance, woody biomass increase for this species was higher in a riverine forest than in a fringe in Terminos Lagoon (Day et al., 1987). Trees occurring in the basin forest of the study site had a much lower woody biomass increase. These results indicate increasingly stressful conditions from riverine forests to basin forests. Thus, environmental conditions such as nutrient input, soil salinity and tidal activity, which define and characterize mangrove forests in terms of litter production, are also important in affecting woody growth rates (Lugo and Snedaker, 1974, Lugo et al., 1988). Therefore, the productivity pattern riverine > fringe > basin also fits well when considering woody growth rates for the same mangrove species. 4.4. Total above-ground net primary production Annual average estimates of total above-ground net primary production (NPP) in the basin forest are lower than fringe and riverine forests (Day et al., 1987; Christensen, 1978; Table 3). For instance, riverine forests in Sri Lanka and Terminos Lagoon had values of 2415 and 2457 g m -2 year-l, respectively (Day et al., 1987; Amarasinghe and Balasubramaniam, 1992). Rhizophora spp. dominated fringe forests in Puerto Rico, Sri Lanka and Terminos Lagoon had high values of 1007, 1388 and 1606 g m - 2 year- L, respectively. The NPP of an A. marina basin forest in Australia was 518 g m -2 year- 1 (Clough and Attiwill, 1982). This value is similar to that of zone II where environmental conditions, in terms of hydrology, nutrient inputs and soil salinity, resemble those of a scrub forest. Mangrove forests flushed frequently by tides and exposed to high nutrient concentrations have higher net primary production (Day et al., 1989). The coefficient of variation of annual NPP was similar to the variability of NPP reported by Putz and Chan (1986) in a Malaysian mangrove forest. The mean stem production:litter fall ratio for the basin forest study site at Estero Pargo (0.4) was lower than those for fringe (0.8) and riverine (1.1) forests in Laguna de Terminos calculated from productivity values reported by Day et al. (1987). By comparison, Lugo et al. (1988) reported ratios of 1.22 and 0.6, respectively, for riverine and basin freshwater wetland forests. The decreasing ratio value from riverine to basin forests indicates that there is relatively less wood production per unit litter fall with increasing stress due to soil salinity, decreasing nutrient inputs and flushing. The significant inverse relationship between litter fall and soil salinity is an indication of this (Fig. 4).

J.W. Day Jr. et al./Aquatic Botany 55 (1996) 39-60

57

4.5. Factors affecting inter-annual productivity patterns Climatic factors such as radiation, day length, air temperature, rainfall, potential evapotranspiration, as well as their seasonal variability play an important role in affecting net primary production and growth of mangrove forests around the world (Clough, 1993). Seventy-four percent of the total litter fall variability over the 7-year study period was explained by changes in average soil salinity, minimum temperature in the norte season, and minimum rainfall in the dry season. Long-term litter fall (32 months) in more species-rich mangroves of Australia was strongly related to annual rainfall (Williams et al., 1981). Litter fall in a mixed basin forest in Florida over an 8 year period averaged 709 g m -2 year- ~ with a coefficient of variation of 10.9% (Pool et al., 1975; Twilley et al., 1986). This is similar to the variation in our study. The coefficient of variation ( C V ) over the study period can be used as an index of the degree of inter-annual variability of aspects of the productivity and litter dynamics of the basin forest at Estero Pargo. The CV of annual litter fall, wood growth per tree, total wood production and NPP in zone II was lower than corresponding values for zones I and III, suggesting that the stressful conditions in zone II lead not only to lower but more variable productivity. The CV of individual productivity measures were higher than those of NPP. The mean CV of productivity measures was less than those of litter standing crop and litter turnover. Previous studies have attempted to develop equations relating global patterns of litter fall production to climate patterns. Leith (1975) found that temperature and rainfall could be used independently to predict net primary productivity of temperate terrestrial ecosystems, but few tropical forest were included in this analysis. Brown and Lugo (1982) used the ratio of annual air temperature to annual precipitation as an index of potential water availability in exploring the relationship with tropical forest litter fall production. This index showed that tropical moist forests are more productive than forests occurring in the driest and hottest tropical regions because of water availability. Studies of litter fall production in differing mangrove forests have shown that mangrove trees drop their leaves throughout the year, although higher litter fall rates have been observed at the end of the dry season and at the onset of the wet season (Pool et al., 1975; Williams et al., 1981; Twilley et al., 1986; Flores-Verdugo et al., 1987). Lopez-Portillo and Ezcurra (1985) found that litter fall was higher in an A. germinans dominated basin forest during the dry season and was directly associated (R 2 = 0.65) with water level, evaporation, temperature and radiation but precipitation was not an important climatic factor in explaining litter fall variability. This lack of correlation between the growth rate of A. marina and rainfall has been reported in other studies (Wium-Anderson and Christensen, 1978), and Lopez-Portillo and Ezcurra (1985) suggest that the different responses of Avicennia are associated with the characteristic geomorphic environment and location along the intertidal zone that it occupies. Our results suggest that environmental conditions occurring during the dry season lead to decreased soil moisture in the forest floor due to lack of precipitation and very low tides, and high evapotranspiration due to high temperatures. This imposes increasing water stress on mangrove trees so that leaf senescence is enhanced, causing leaf fall rates to be highest during this season. Similar results have been reported for Avicennia

58

J.W. Day Jr. et al./ Aquatic Botany 55 (1996) 39-60

sp. in mangrove forests in Australia (Duke et al., 1981; Williams et al., 1981) and in upland tropical forests (Reich and Borchert, 1984; Wright and Cornejo, 1990) where peaks in leaf fall occurred during the dry season when water stress was considered an important factor in causing high litter fall. High litter fall in our site during the rainy season, when water availability was not an important factor, was partially related to increased flower and fruit fall (53% of the total non-leaf material fell during this season compared with 27% and 20% during the dry and norte seasons, respectively), as has been reported for the Laguna de T6rminos region and elsewhere (Lugo and Snedaker, 1974; Christensen, 1978; Day et al., 1987). It has been suggested that the fall of flowers and fruits during the wet season is of adaptive value for mangrove trees growing in flooded environments since it permits the fruits and seeds to be dispersed during high water level conditions that usually occur during the rainy season (Rabinowitz, 1978; Jimenez and Sauter, 1991). However, further studies are needed to clearly understand not only the factors that regulate fruit and seed dispersion, but also the factors that regulate seed establishment, survival and growth (Smith, 1993). It seems that mangrove trees allocate more energy to growth during this season when environmental conditions, particularly low salinity and enough freshwater coming from precipitation and runoff, are suitable for plant growth so that leaf fall is likely reduced for optimizing photosynthesis.

Acknowledgements This paper is the result of a cooperative research program sponsored by the Instituto de Ciencias del Mar y Limnolog~a, Universidad Nacional Aut6noma de M6xico, the EPOMEX program, Universidad Aut6noma de Campeche and Louisiana State University. We thank the staff of the marine laboratory in Cd. del Carmen for logistical support especially Maria del Carmen Ramfrez and Juan Luis Casanova. Support for this study was provided by the US Agency for International Development, the US Man and Biosphere Program, the International Union for the Conservation of Nature and respective participating institutions.

References Amarasinghe, M.D. and Balasubramaniam, S., 1992. Net primary productivity of two mangrove forest stands on the northwestern coast of Sri Lanka. Hydrobiologia, 247: 37-47. Boto, K.G., Bunt, J.S. and Wellington, J.T., 1984. Variations in mangrove forest productivity in northern Australia and Papua New Guinea. Estuarine Coastal Shelf Sci., 19: 321-329. Brown, S. and Lugo, A.E., 1982. A comparison of structural and functional characteristics of saltwater and freshwater wetlands. In: B. Gopal, R. Turner and R. Wetzel (Editors), Wetlands Ecology and Management. International Scientific Publishers, Jaipur, India, pp. 109-130. Christensen, B., 1978. Biomass and primary production of Rhizophora apiculata Bl. in a mangrove forest in southern Thailand. Aquat. Bot., 4: 43-52. Clough, B.F., 1993. Primary productivity and growth of mangrove forests. In: A.I. Robertson and D.M. Alongi (Editors), Tropical Mangrove Ecosystems. Australian Institute of Marine Science and American Geophysical Union, Washington, D.C., pp. 225-249.

J. W. Day Jr. et al. / Aquatic Botany 55 (1996) 3 9 - 6 0

59

Clough, B.F. and Attiwill, P.M., 1982. Primary productivity of mangroves. In: B.F. Clough (Editor), Mangrove Ecosystems in Australia. Books Australia, Miami, FL, pp. 213-222. Day, J.W., Conner, W.H., Ley-Lou, F., Day, R.H. and Machado, A.N., 1987. The productivity and composition of mangrove forests, Laguna de T6rminos, M6xico. Aquat. Bot., 27: 267-284. Day, J.W., Hall, C., Kemp, M. and Y(uSez-Arancibia, A., 1989. Estuarine Ecology. Wiley lnterscience, New York, 558 pp. Duke, N.C., Bunt, J.S. and Williams, W.T., 1981. Mangrove litter fall in north-eastern Australia. I. Annual totals by component in selected species. Aust. J. Bot., 29: 547-553. Flores-Verdugo, F.J., Day, J.W. and Brisefio-Duefias, R., 1987. Structure, litterfall, decomposition, and detritus dynamics of mangroves in a Mexican coastal lagoon with ephemeral inlet. Mar. Ecol. Prog. Ser.. 35:51-56. Goulter, P.F.E. and Allaway, W.G., 1979. Litter fall and decomposition in a mangrove stand. Avicennia marina (Forsk.) Vier., in Middle Harbour, Sydney. Aust. J. Freshwater Res., 30: 541-546. Jimenez, J.A. and Sauter, K., 1991. Structure and dynamics of mangrove forests along a flooding gradient. Estuaries, 14: 49-56. Leach, G.J. and Burgin, S., 1985. Litter production and seasonality of mangroves in Papua New Guinea. Aquat. Bot., 23: 215-224, Lee. S.Y., 1989. Litter production and turnover of the mangrove Kandelia candel (L.) Druce in a Hong Kong tidal shrimp pond. Estuarine Coastal Shelf Sci., 29: 75-87. Leith, H., 1975. Modeling the primary productivity of the world. In: H. Leith and R.H. Whittaker (Editors), Primary Productivity of the Biosphere. Springer, New York, pp. 237-263. Lopez-Portillo, J. and Ezcurra, E., 1985. Litter fall of Avicennia germinans L. in a one-year cycle in a mudflat at the Laguna de Mecoac~n, Tabasco, Mexico. Biotropica, 17: 186-190. Lugo, A.E. and Snedaker, S.C., 1974. The ecology of mangroves. Annu. Rev. Ecol. Syst., 5: 39-64. Lugo, A.E., Brown, S. and Brinson, M., 1988. Forested wetlands in freshwater and salt-water environments. Limnol. Oceanogr., 33(4, part 2): 894-909. Lynch, J.C., Meriwether, J.R., McKee, B.A., Vera-Herrera, F. and Twilley, R.R., 1989. Recent accretion in mangrove ecosystems based on 137Cs and 2~°Pb. Estuaries, 12:284 299. Olson, J.S., 1963. Energy storage and the balance of producers and decomposers in ecological systems. Ecology, 44: 322-331. Onuf, C.P., Teal, J.M. and Valiela, I., 1977. Interactions of nutrients, plant growth, and herbivory in a mangrove ecosystem. Ecology, 58:514-526. Pool, D.J., Lugo, A.E. and Snedaker, S.C., 1975. Litter production in mangrove forests of southern Florida and Puerto Rico. In: G.E. Walsh, S.C. Snedaker and H.J. Teas (Editors), Proc. of the Int. Syrup. of Biology and Management of Mangroves, Vol. 1, University of Florida, Gainesville, FL, pp. 213-237. Pool, D.J., Snedaker, S.C. and Lugo, A.E., 1977. Structure of mangrove forests in Florida, Puerto Rico, Mexico, and Costa Rica. Biotropica, 9: 195-212. Putz, F.E. and Chan, H.T., 1986. Tree growth, dynamics, and productivity in a mature mangrove forest in Malaysia. For. Ecol. Manage., 17:211-230. Rabinowi~, D., 1978. Early growth of mangrove seedlings in Panama, and hypothesis concerning the relationship of dispersal and zonation. J. Biogeogr., 5:113-133. Reich, P.B. and Borchert, R., 1984. Water stress and tree phenology in a tropical dry forest in the lowlands of Costa Rica. J. Ecol., 72: 61-74. Rico-Gray, V. and Lot, A., 1983. Producci6n de hojarasca del manglar de la Laguna de la Mancha, Veracruz, M6xico. Biotica, 8: 295-301. Rivera-Monroy, V., Day, J.W., Jr., Twilley, R., Vera-Herrera, F. and Coronado-Molina, C., 1995. Nitrogen exchange between a fringe mangrove forest and the adjacent tidal creek and the influence on aquatic primary production. Estuarine Coastal Shelf Sci., 40:139-160. Robertson, A.1. and Alongi, D.M. (Editors), 1993. Tropical Mangrove Ecosystems. Australian Institute of Marine Science and American Geophysical Union, Washington, D.C., 321 pp. Smith 11I, T.J., 1993. Forest structure. In: A.I. Robertson and DM. Alongi (Editors), Tropical Mangrove Ecosystems. Australian Institute of Marine Science and American Geophysical Union, Washington, D.C., 329 pp. Statistical Analysis Systems Institute Inc., 1991. SAS User's Guide: Statistics. SAS Institute Inc.. Cary, NC.

60

J.W. Day Jr. et al./ Aquatic Botany 55 (1996) 39-60

Thom, B.G., 1982. Mangrove ecology--A geomorphological perspective. In: B.F. Clough (Editor), Mangrove Ecosystems in Australia. Canberra Australian National University Press, Canberra, pp. 3-17. Twilley, R.R., Lugo, A.L. and Patterson-Zucca, C., 1986. Litter production and turnover in basin mangrove forests in Southwest Florida. Ecology, 67(3): 670-683. Williams, W.T., Bunt, J.S. and Duke, N.C., 1981. Mangrove litter fall in northeastern Australia II. Periodicity. Aust. J. Bot., 29: 555-563. Wium-Anderson, S. and Christensen, B., 1978. Seasonal growth of mangrove trees in southern Thailand. II. Phenology of Bruguiera cylindrica, Ceriops tagal, Luminifera littorea and Avicennia marina. Aquat. Bot., 5: 383-390. Wright, J.S. and Cornejo, F.H., 1990. Seasonal drought and leaf fall in a tropical forest. Ecology, 7(3): 1165-1175.

Yfifiez-Arancibia, A. and Day, J.W., Jr., 1982. Ecological characterization of T6rminos lagoon, a tropical estuary in the southern Gulf of Mexico. Oceanol. Acta, SP: 431-440.